In [1]:

    

import pandas as pd
import numpy as np


    

In [2]:

    

df = pd.read_csv('../data/M4Comp.csv.gz')


    

In [3]:

    

df.head()


    

    
Out[3]:






  

    

      

      
K
      
ID
      
TYPE
      
H
      
PERIOD
      
UNITS
      
N
      
PAST
      
FUTURE
    
  
  

    

      
0
      
SERIES
      
FIN1
      
FINANCE
      
8
      
MONTHLY
      
NaN
      
170
      
87.33542282,86.10394147,84.98490317,84.4860577...
      
85.00849262,82.86804888,83.14415073,83.1855927...
    
    

      
1
      
SERIES
      
FIN2
      
FINANCE
      
14
      
MONTHLY
      
NaN
      
331
      
111.9,109.8,109.5,108.8,117.9,122.8,120.9,119....
      
285.9,288.9,283.2,293.4,291.2,301,302.1,317.2,...
    
    

      
2
      
SERIES
      
FIN3
      
FINANCE
      
14
      
MONTHLY
      
NaN
      
331
      
91.9,89.5,89.8,91.5,99.1,102.9,100.5,99.1,102....
      
96.4,96.7,94.5,97,96.3,98.6,98.8,102.8,100.7,9...
    
    

      
3
      
SERIES
      
FIN4
      
FINANCE
      
8
      
MONTHLY
      
NaN
      
178
      
3.949,3.793,3.611,3.597,3.51,3.327,3.482,3.446...
      
1.48,1.718,1.52,1.334,1.448,1.173,1.276,1.406
    
    

      
4
      
SERIES
      
FIN5
      
FINANCE
      
8
      
MONTHLY
      
NaN
      
178
      
4.39,4.14,3.828,3.871,3.811,3.859,4.04,3.939,3...
      
1.624,1.662,1.737,1.591,1.629,1.592,1.461,1.509
    
  

In [4]:

    

df.PERIOD.value_counts()


    

    
Out[4]:





MONTHLY       3750
YEARLY        2240
WEEKLY        1833
DAILY         1762
QUARTERLY      402
BIANNUALLY      13
Name: PERIOD, dtype: int64

In [5]:

    

df.TYPE.value_counts()


    

    
Out[5]:





ECONOMICS            2000
FINANCE              2000
INVENTORY            1500
BUSINESS-INDUSTRY    1500
DEMOGRAPHICS         1000
CLIMATE              1000
INTERNET-TELECOM     1000
Name: TYPE, dtype: int64

In [6]:

    

lDict = df.TYPE.value_counts().to_dict()


    

In [9]:

    

for ty in lDict.keys():
    df[df.TYPE == ty].to_csv('../data/M4Comp_' + ty + ".csv")


    

In [9]:

    

df[df.H > 100].head()


    

    
Out[9]:






  

    

      

      
K
      
ID
      
TYPE
      
H
      
PERIOD
      
UNITS
      
N
      
PAST
      
FUTURE
    
  
  

    

      
26
      
SERIES
      
FIN27
      
FINANCE
      
127
      
DAILY
      
NaN
      
3514
      
0.844,0.845,0.845,0.843,0.84,0.844,0.847,0.851...
      
2.3752,2.375,2.3727,2.3975,2.3291,2.313,2.323,...
    
    

      
27
      
SERIES
      
FIN28
      
FINANCE
      
328
      
DAILY
      
NaN
      
9136
      
1.0109,1.0102,1.0106,1.0148,1.0154,1.0159,1.01...
      
1.0854,1.0816,1.0841,1.0794,1.0719,1.0753,1.07...
    
    

      
28
      
SERIES
      
FIN29
      
FINANCE
      
242
      
DAILY
      
NaN
      
6742
      
1.5341,1.5418,1.5264,1.5264,1.5264,1.5371,1.53...
      
7.2115,7.195,7.1965,7.1912,7.1818,7.1845,7.191...
    
  

In [ ]: